Liu 1995.

Methods	RCT Approved by the ethics committee and informed consent obtained Setting: United States of America Funding: charity (plus a contribution from industry)
Participants	54 ASA 1 to 3 patients undergoing elective partial resection of the colon
Interventions	Treatment group: TEA (T8‐T9 or T9‐T10; catheter threaded 3 cm) with bupivacaine 0.15%, 10 mL/h for various times (n = 14) or combinations of epidural morphine 0.03 mg/mL + bupivacaine 0.1%, 10 mL/h for various times (n = 14) Control group: postoperative epidural morphine 0.05 mg/mL 10 mL/h for various times (n = 12) Analgesia at rest was titrated to a verbal pain score < 5/10 (0 = no pain, 10 = worst possible pain) with an epidural injection of fentanyl 50 micrograms followed by an increase in the epidural infusion of 2 mL/h every hour as needed. Therefore, all participants received epidural opioids, and the 2 groups with bupivacaine were fused to be compared with the control group Length of epidural infusion taken as ≥ 48 hours, although not clearly stipulated
Outcomes	Time of first flatus VAS scores at movement at 24, 48 and 72 hours Hospital LOS Anastomotic leakage
Notes	One group with postoperative IV PCA morphine not retained for analysis Groups B and MB were ready for discharge an average of 35 hours earlier (95% confidence interval for difference ranges from 27 to 49 hours) than groups M and P
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Separate randomization tables were prepared for each institution. Randomization was stratified by planned left vs right colonic anastomosis because left colonic anastomosis may result in greater postoperative ileus
Allocation concealment (selection bias)	Unclear risk	Unclear (randomized on arrival to the preoperative holding area)
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Epidural groups (MB, M and B) were double‐blinded
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Epidural groups (MB, M and B) were double‐blinded
Incomplete outcome data (attrition bias) All outcomes	Low risk	No other loss to follow‐up (apart from the 2 withdrawn because a catheter could not be inserted)
Selective reporting (reporting bias)	Low risk	All results reported
Other bias	Unclear risk	Groups well balanced Not in intention‐to‐treat: Epidural catheters could not be placed in 2 participants, who were withdrawn from the protocol